Fully Convolutional Networks for

Introduction

We show that a fully convolutional network (FCN), trained end-to-end, pixels-to-pixels on semantic segmentation exceeds the state-of-the-art without further machinery.

Related work

Fully convolutional networks
Dense prediction with convnets

Fully convolutional networks

Adapting classifiers for dense prediction
Shift-and-stitch is filter rarefaction
Upsampling is backwards strided convolution
Patch wise training is loss sampling

Segmentation Architecture

We cast ILSVRC classifiers into FCNs and augment them for dense prediction with in-network upsampling and a pixelwise loss. We train for segmentation by fine-tuning. Next, we build a novel skip architecture that combines coarse, semantic and local, appearance information to refine prediction.

因为32stride的缩减影响了分割的精度
所以用跳跃连接的方式来补充细节信息

Learning this skip net improves performance on the validation set by 3.0 mean IU to 62.4.

Experimentalframework

Optimization
momentum 0.9, weight decay of 5−4 or 2−4

Fine-tune
全网络调优

Path Sampling
We study this tradeoff by spatially sampling the loss in the manner described earlier

Class Blancing
Fully convolutional training can balance classes by weighting or sampling the loss

Dense Prediction

Augmentation
镜像等方式增加数据

More Training Data
使用了更多的训练数据

Results